Discussion Forum

Hi

from my experience you need some air around the fibre, then a PML of air all around (basically the PML should have the same material as the adjacent domain, and should all be the same.

I do believe it will work to extend the glass and have PML in glass, but that means you have an infinite extend of glass and no glass air interface, but pls check carefully if this works correctly, I'm not 100% sure

--
Good luck
Ivar

Akhilesh,

In a conventional step index fiber the modes inside the core are so confined that air has a negligible role in the simulations. If that is your case, you may put the PMLs over the cladding (that means the real part of the PML's refractive index shall be that of the silica). I recommend to put the absorbing walls a bit far from the core (several lambdas) so to avoid spurious fields inside the PML.

Cheers,
--
Felipe

Thanks to both Ivar and Felipe,

I am doing simulation of the NEMO fiber( I have attached the SEM image from Geernert et.al. IEEE photonics letters,20, 8, 2008) with dooped core( defect in the centre).

I have put a rectangular PML around the airholes so as to enclose the microstructure,
If I give assign the PML refractive index the cladding refractive index I do not get fundamental mode at the expected mode index but if I do the same with air refractive index I get the results.

I do not know exactly as to why the air refractive index for PML works even though its inside the cladding. Can someone clear my doubt?

Akhilesh

Hi

I understand that as the effect of the glass air interface has it's importance for the overall EM field. I would anyhow better trust a model that includes the overall environment, hence air, and use the infinite or PML layer sufficiently far away from the high index structure so that any evanescent wave has mainly died out before one enter the PML layer

--
Good luck
Ivar

The purpose of PML is to reduce the computational domain and if we place it far away does'nt it loose its purpose?

Hi

Indeed so it's our definition of "far" that is not in line.

For me it means a few wavelength as a minimum, just to be sure that whatever is reflected (even if it should not be) is not directly lined up to give too high interferences, so in now way even cm's, for a standard fibre model

--
Good luck
Ivar

Hi,
Thanks Mr. Ivar for your help.

I am using cartesian PML in version 4.2a and I do not know how to set the absorbing directions for it.
It also gives me error about internal and external boundary condition.

Please help.

Hi

if your shape is a square then cartesian is good, but try to center it w.r.t the device and the PML centre, the same if you choose a circle or a sphere in 3D, it's important to set the centre position accordingly

--
Good luck
Ivar

Hi,

In the previous comsol versions like 3.5 one could choose the absorbing directions for the PML but in version 4.2a I cant find anything that can do that.

If anyone has any idea please reply.

Thanks
Akhilesh

you don't need to do it in 4.2a. You only need to select type of symmetry (rectangular, cylindrical or spherical). COMSOL takes care of the rest.

For example, if you have incident plane wave, you don't need specify the angle. It will be absorbed at all angles.

Hi

thaks, that is good to know ;)

--
Good luck
Ivar

Hi,
Thanks Alexander,
But I get the following error when I choose cartesian PML
"Automatic scaled regions cannot treat selections without exterior and interior boundaries"

I dont know how to get rid of the error.

Looks like there are problems with the geometry and assignment of PMLs. Upload your model so we can have a look. It's hard to say what's exactly wrong without seeing the geometry.

I have the same problem. I am getting this error "Automatic scaled regions cannot treat selections without exterior and interior boundaries." Does anyone have any idea?

Hi

I believ it means that a PML or INF "layer" must be an external domain surrounding your model

--
Good luck
Ivar

Hello Sir, I encountered a problem with the value of the effective index of the modes that are equal (effective index of the mode 1 = effective index mode 2) please commons to solve this problem. help me please

Hi,
I have been experiencing the same error message while computing. Could you please help me understand the problem that I am facing and the solution for it.

Thanks.

Dear Ivar,

I have a problem related to PML refractive index. I am simulating the scattering cross sections of a gold nano disk (diameter 100 nm and height 20 nm) around homogeneous media. I use a PML layer as a block with 1000nmX1000nmX1000nm dimension and 200 nm thickness. I mean the width,depth and height of PML is 1000+(2X200) i.e. internal block with 1000nm dimensions and external block with 1400 nm dimensions and in between 200 nm thickness in two sides. When the whole surrounding and PML (internal and external blocks) is Air, I am getting scattering cross sections and fields correctly, no reflection at PML boundary and my results closely matches with reference research papers values. So I am confident that my simulation not going wrong. But when I change the material of the PML and surrounding (internal and external blocks) as glass with n=1.5, I am getting scattering cross sections and fields wrong, and reflection at PML boundary. More strange is, for 400nm the reflection at PML boundary is very strong (RED color in 3D color plot of field) but for 800 nm there is no reflection. Now My question is , shouldn't be all domains except the gold nano disk is glass to create the homogeneous surrounding around disk? Like in the case of air it works and why for glass it will not? Or, The PML will be air around the glass and gold disk inside the glass? which is the correct way to simulate this problem. I choose scattering boundary conditions at Internal PML surfaces.